The EMILIO project, a European initiative, seeks to address these challenges by developing smart home monitoring systems integrated with wearable devices to track daily activities and movement patterns. By analyzing the collected data using machine learning algorithms, these systems can detect anomalies, such as deviations from regular routines, and send alerts to caregivers or family members, allowing for timely intervention and a better understanding of potential health risks

Nathan Vandemoortele from IDLab's PreDiCT team, tested this concept in HomeLab, where ultra-wideband (UWB) sensors were integrated as a novel contextual sensing method to accurately track a person’s position within a room. Combined with accelerometer data from wearable devices, this setup allows machine learning models to predict daily activities and identify deviations from routine. Such detailed monitoring helps detect subtle behavioral changes that may indicate emerging health concerns, such as missed meals, unusual sleep patterns, or episodes of reduced mobility.

Traditionally, elderly care relied on direct supervision or periodic check-ins. By using HomeLab as a testbed for smart monitoring technologies, we refined our AI-driven models and sensor integration. This research paved the way for real-time, non-intrusive monitoring in a real-world setting at Vulpia, where our system provides caregivers with real-time insights into residents’ well-being. By detecting early signs of potential health issues, caregivers can take proactive measures, ultimately improving the quality of life of our seniors.

Calibration of the UWB positioning system at HomeLab

Activity of Daily Living (ADL) monitoring at Vulpia (Lilian and Bob). Large Language Models have become state of the art at interpreting semantically rich context and zero-shot classification of activities of daily living